If the package was linked to Ring1 from Ring1, staging list command
should display it as Ring0 package ie. it's only be able to puts in
ring0 staging project. This is work out
https://progress.opensuse.org/issues/8266
With --split argument, staging adi command can splits each package to
different adi staging project, the argument can work by 'osc staging adi
--split' or 'osc staging adi --move --split'. Note that, it will be
ignore --by-develproject argument too.
Update by_dp function in adi command, the old way is incompatible to
newer Leap development model, now we should try to check target package
in Factory as well, otherwise the submission from any other place than
Factory did not have 'devel' entry in _meta.
Workaround glibc.i686 package config, it had a topadd block in _link,
and looks like it causes the disturl won't consistently with glibc even
with the same srcmd5. And the lastsuccess state will be outdated if the
revision is used the same srcmd5.
Note: the proper workaround, to instead of
5daf2bb6509abaf966268c9d6b13aa4e330a8af3
Workaround glibc.i686 package config, it had a topadd block in _link,
and looks like it causes the disturl won't consistently with glibc even
with the same srcmd5. And the lastsuccess state will be outdated if the
revision is used the same srcmd5.
Add glibc.i686 to specs if its glibc request, therefore, repochecker
will not ignore glibc.i686 binaries in case some binaries need them.
Also ignore build_excluded if it's i686 only package, and downloading
it's i586 packages.
The additional message will look more clear than telling it's bad
disturl and no matching downloads available. Note that, before executing
in check_disturl(), there is another place to check the revision which
was in check_specs(), but it was mainly for 2nd spec.
And makes DECLINED message consistently.
Some users are confused as the old message "package never built" does
not have to be true in many cases, especially on longer-lasting
stagings. In this case, the binaries might have been built but have
since been replaced by newer ones - the effect is the same: repo-checker
can't receive them in order to perform the full checks.
If release-notes-sles is existed then it is not openSUSE Factory but
SLE, hence do not check totest and openSUSE-release.rpm on it.
And add try/except protection in get_binary_version otherwise it will
crashes in case the package is not existed.
Adding the common Factory build repository to the candidate repos only,
adn re-introduced build_excluded attribute. If it haven't good repo ie.
builds against ARM, PPC, images, i586 only or whatever, leave them there
for human to check.
Verified with SR#359327, SR#360767, SR#361364 and SR#361243.
